Calculating Alpha, Beta, Delta, Theta w. Emotiv

[roschler]

I want to calculate the Alpha, Beta, Delta, Theta spectral bands with the Emotiv headset. I was told to look at the temporal filter code as a starting point. While searching through the code I came upon a source file related to the Neurosky headset:

ovasCDriverNeuroskyMindset.cpp

In it there are references to the spectral bands. I'm hoping to save a little time so if someone could give me some pointers on where to look/what to do to calculate the same measurements using the Emotiv headset, I'd appreciate that. If not I'll scour the source code chain from that file and figure it out myself, but I thought maybe someone else had tried calculating those values for the Emotiv headset and good give me some starting points, especially as to where the Neurosky code is doing that so I can pull from that section of the trunk.

-- roschler

[jlegeny]

Hello roschler,

there is actually no need to create any additional code. OpenViBE already has all of the necessary components to do the calculations by itself. You should use the temporal filter, as suggested. The only necessary change is to set the filter to act as a band-pass filter for the spectral bands of your choice. For example if you want to calculate the alpha band you will set the low cut-off frequency to 8 and the hight cut-off to 13 (approximately, the definition of what exact values to use are up to your research). You will probably want to do some spatial filtering (again, there is a box for it) beforehand to eliminate as much noise as possible. Finally, you will need one temporal filter box per frequency band you want to detect.

Cheers
Jozef

[hatma]

Dear all

I still confuse about how to remove noise on EEG signals.
currently I am using EMOTIV device.

does it mean Spatial Filter box will remove noise?
I thought it only generates a number of output channels from another number of input channels?

sorry for my lack knowledge about EEG signal processing.

best regard
hatma

[fabien.lotte]

Dear Hatma,

Removing noise from EEG signals is quite a tricky problem actually. The spatial filter box is one of the tools to do so, because it enables us to define a weighted combination of sensors which contain the noise and thus can be used to remove it from the EEG signals. The question is how to define this weighted combination. For instance, you can use ICA (Independent Component Analysis) to obtain this combination and identify the components which look like noise. Or you can use classification oriented optimization algorithm for spatial filters such as CSP or xDAWN. In any case, there is no generic and simple way to remove noise, it is still an open signal processing problem

I hope this helps,
Fabien